ADR-0006 D1 retirement addendum: record that D2 executed and the three retained-project surfaces are renamed

[hotlong]

Part of #12865 (epic parent; unparked 2026-08-28 by maintainer ruling recorded there). Filed by epic PM session session_65d2faee-1ff4-4be8-be28-72b972c539d9.

Scope

Append a retirement addendum to docs/adr/0006-project-environment-split.v4.md recording that the D2 pre-registration fired: the maintainer's 2026-08-28 ruling on #12865 opened the window, and the three retained-project surfaces named by D1 (the client SDK projects namespace, the response envelope keys, the JSDoc sentence) are renamed by the executing PRs (link the SDK-half PR from #12866 and the producer-half PR from objectstack-ai/cloud#1691 once they exist). D1 is retired — retire, never alias, per D2's own instruction.

Follow the MEASURED addendum conventions PR #12736 established and verified (top-level dated addendum heading appended after References, Status line amended to point at it, a why-an-addendum-not-a-new-ADR paragraph). Re-measure the file's state at branch time — do not assume this card's description.

Blocked-by: #12473

(That card's PR #12736 authored the D1/D2/D3 addendum this card retires. PR #12736 is an open DRAFT on the governed surface awaiting HUMAN merge. This card cannot start until it lands — starting earlier edits the same file against an unmerged base.)

Landing constraint

Governed surface (docs/adr): draft-only PR, human merge, never auto-merge. Belongs to the same coordinated release as the rename halves. skip-changeset likely correct (doc-only) but derive it, do not copy.

[hotlong]

Claim: PM loop round 3 (epic #12865, maintainer-directed; blocking edge resolved — PR #12736 human-merged, #12473 closed)
Session: session_65d2faee-1ff4-4be8-be28-72b972c539d9
Branch: claude/issue-12867-adr0006-d1-retirement-addendum
Worktree: objectstack-issue-12867
Domain: domain:skills per triage label; dispatched via the epic subtree (epic territory declared on #12865)
File surface: docs/adr/0006-project-environment-split.v4.md only (stop on breach; explain in the report)
Container & model: M docs-judgment, mode:subagent, model: opus (tier probe precedent: no path-derived mandate for docs/adr; PM judgment)
Clause-②: no
(Docs-only ADR addendum; changes no contract accept/reject behavior and widens no public surface — it records an executed, thrice-ruled rename.)
Serial constraints cleared: re-derived this dispatch across every open PR's file list — zero open PRs touch docs/adr/0006*. Predecessor PR #12736 (same file) is MERGED; branch from origin/main at or after its merge. Read-coupling siblings: PR #12885 and cloud PR #1692 (the executing rename PRs this addendum cites) are ACCEPTed drafts parked for the coordinated window — cite them as authored-and-parked, not as released.

[hotlong]

Claim: os-dev seat, dispatched by the epic PM (round 3) for this card and no other.

Session: session_65d2faee-1ff4-4be8-be28-72b972c539d9 (dev subagent under the epic PM session; the seat shares the dispatching session id, so the identity bit that distinguishes this run is the branch below, not the session string)
Branch: claude/issue-12867-adr0006-d1-retirement-addendum
Worktree: objectstack-issue-12867, cut from fresh origin/main at 0c2334f6c
Base check: PR #12736's merge commit ac35e2a2a7f8c05e28cfab50dd6a0b0406adfdcf verified as an ancestor of this branch point — the D1/D2/D3 addendum this card retires IS on the base. The blocking edge on #12473 is resolved.
File surface: docs/adr/0006-project-environment-split.v4.md only. No changeset FILE (the changeset gate is satisfied by the skip-changeset label, derived from pr-automation.yml and matched to the measured precedent of PR #12736, which touched this same file alone and carried that label).
Landing constraint acknowledged: governed surface — DRAFT PR, human merge, never un-draft, never arm auto-merge; rides the same coordinated release as PR #12885 and objectstack-ai/cloud#1692.

Empty branch pushed before the first edit as a write-route probe: succeeded.

---

Generated by Claude Code

[hotlong]

os-dev-report

{
  "issue": 12867,
  "status": "done",
  "branch": "claude/issue-12867-adr0006-d1-retirement-addendum",
  "pr": "https://github.com/objectstack-ai/objectstack/pull/12908",
  "premise_still_valid": true,
  "files_changed": ["docs/adr/0006-project-environment-split.v4.md"],
  "summary": "Appended a second dated addendum to ADR-0006 v4 recording that D2 fired on the maintainer's 2026-08-28 epic ruling and that D1's three deliberately-retained `project` surfaces are renamed with no aliases. Three edits, one file, +131/-1: the new addendum after the existing one; the Status line EXTENDED with a D1-retired pointer (the ADR-0053 multi-pointer pattern, not a rewrite); and a one-line RETIRED block quote under the D1 heading (the ADR-0025 RETIRED-marker shape) leaving D1's body verbatim so the record of why the surfaces were retained stays legible. All three surfaces cited by QUOTED PHRASE, never line number. Every claim was measured against the two executing PRs' diffs rather than copied from the card, and cited as authored-and-parked, not released.",
  "tests": "Gate union ran at commit 469006493 (final commit; tree clean, remote in sync). Derived with `node scripts/pm/dispatch-gates.mjs --repo objectstack-ai/objectstack` on the actual changed path -- 5 families, two MORE than the dispatch anticipated (it added check:pm-governed-merges and check:doc-formula-expressions). Exit codes captured before any pipe, into per-gate log files; verdicts quoted from each gate's own output. GREEN: check:doc-authoring EXIT=0 ('doc authoring guard: 392 files clean -- no bare metadata literals'); check:pm-governed-merges EXIT=0 ('check-governed-merges --self-test: 206 assertions'); check-adr-links.mjs EXIT=0 ('check-adr-links: 571 relative link destination(s) under docs/adr/ resolve'); check:doc-formula-expressions EXIT=0 ('22 record-scoped formula example(s) across 425 files / 1453 TS blocks judged clean'); check:nul-bytes EXIT=0 ('scanned 7176 text file(s) ... no raw ASCII control bytes'). Also ran check:partof-closing-keyword on the PR body: EXIT=0. NOT MEASURED then resolved: check:doc-formula-expressions first exited 1 twice with 'PREREQUISITE NOT MET -- the workspace package is not built' (formula, then lint); that is not a red, it measured nothing. Built both closures with turbo, then it went green. RED but PRE-EXISTING: check:adr-anchors EXIT=1, 1 failure of 106 assertions -- 'ADR-0006 D4 is cited by 1 file(s), but ADR-0006 declares no D4'. Proven not mine by a positive control: committed first, then restored origin/main's version of the ADR in place under an absolute-path EXIT/INT/TERM trap, re-ran the scan, and diffed the two outputs -- BYTE-IDENTICAL, so this branch changes the gate's output in no way. Restoration proven by an empty `git diff HEAD` plus `git status --porcelain`, and by my marker string counting 0 during the baseline leg and 1 after restore. The citing file, packages/spec/src/data/object.zod.ts line 1467, is outside this card's declared single-file surface.",
  "gates_run": [
    {"gate": "check:adr-anchors", "exit": 1, "outcome": "RED -- pre-existing on origin/main, byte-identical output with and without this branch's change; not caused by this PR"},
    {"gate": "check:doc-authoring", "exit": 0, "outcome": "green"},
    {"gate": "check:pm-governed-merges", "exit": 0, "outcome": "green"},
    {"gate": "check-adr-links.mjs", "exit": 0, "outcome": "green"},
    {"gate": "check:doc-formula-expressions", "exit": 0, "outcome": "green after building @objectstack/formula and @objectstack/lint; two prior exit-1 runs were PREREQUISITE NOT MET and measured nothing"},
    {"gate": "check:nul-bytes", "exit": 0, "outcome": "green"},
    {"gate": "check:partof-closing-keyword", "exit": 0, "outcome": "green -- run against the PR body before creating the PR"}
  ],
  "deviations": [
    "Verification ran UNLOCKED. scripts/pm/os-verify-lock.sh found no usable flock on this macOS host and printed its own official declaration wording; the entry point ran the commands directly and instructed that the narrowing be declared. It is declared in the PR body verbatim, under its own heading. No serialization guarantee held for this run.",
    "Two gate families beyond the dispatch's list were derived and run (check:pm-governed-merges, check:doc-formula-expressions). The dispatch's gate list was treated as a clue, not a spec, per the re-derivation rule.",
    "Filed the out-of-scope finding as a COMMENT on the existing card #12786 rather than a new issue: the dedup search found that card already names this exact site (ADR-0006 D4, one site, packages/spec/src/data/object.zod.ts). A new card would have been a duplicate. The comment records a genuine escalation, see out_of_scope_findings.",
    "Claim comment posted, but the session id it carries is the dispatching epic PM's (session_65d2faee-...): this dev seat runs as a subagent under that session and has no distinct session string available to it. Said so in the claim comment itself; the branch name is the identity bit that distinguishes this run.",
    "No changeset FILE. Derived from .github/workflows/pr-automation.yml: changeset-check has no path-based exemption, so a doc-only PR needs either the skip-changeset label or an empty changeset. Chose the label -- it also keeps the diff inside the card's declared single-file surface, which a changeset file would have breached. Matches predecessor PR #12736 (same ADR file alone, same label). Label written with the additive REST POST endpoint; read back twice, and it survived the size-labeler's pass. Final label set is documentation/size/m/skip-changeset -- identical to PR #12736's."
  ],
  "mcp_calls": "0 -- REST probe succeeded at the start of the run, so every GitHub read and write went through repo-scoped REST via gh; no MCP GitHub calls were made. Dedup used two targeted /search/issues queries, no wide listing.",
  "open_questions": [],
  "out_of_scope_findings": [
    "Recorded on the existing card #12786 (comment 5450025028) instead of a duplicate new issue: check:adr-anchors is RED on origin/main today and the failure MODE has escalated since that card was measured. #12786 recorded ADR-0006 as declaring no decision letters, so the ADR-0006 D4 citation in packages/spec/src/data/object.zod.ts line 1467 scored a soft 'cannot verify' note. PR #12736 then gave ADR-0006 a real letter index (D1, D1.1, D1.2, D1.3, D2, D3), which flips that same citation into a hard failing assertion -- the gate now exits 1 for every PR on main until the citation is respelled or the decision is written. I deliberately did NOT mint a D4 in this addendum to green it: this card retires D1, it does not decide namespace retirement, and manufacturing a decision letter to satisfy a checker is exactly the defect the gate names. PM may want this in the coordinated window's every-check-green list alongside the #12881 gate item already on the epic checklist."
  ]
}

---

Generated by Claude Code